The main purpose of the study is to build a system to perform associative classification on spatio-temporal sequences. The proposed methodology is composed of four ordered phases: preprocessing, frequent itemsets mining, association rules generation and prediction model training. The model presented is eventually compared to other state-of-the-art classification algorithms such as Decision Trees, Random Forests and Support Vector Machines. On balance, the prediction model achieves a higher precision for the critical and most rare class with respect to its competitors.